> ## Documentation Index
> Fetch the complete documentation index at: https://docs.jaasskills.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Security & trust chain

> How a published artifact earns trust, and how access is enforced

## Trust chain

<Steps>
  <Step title="Build">Artifact generated in a CI pipeline.</Step>
  <Step title="Digest">Digest computed from the package payload.</Step>
  <Step title="Sign">Signature produced through Cosign/Sigstore.</Step>
  <Step title="Verify">Registry verifies the signature against organizational trust policy.</Step>
</Steps>

## Access control

* JWT validation at the request boundary
* Scope checks against the caller's permissions manifest
* Optional tenant boundary enforcement via audience and tenant claims
* Visibility/sharing filter on search and metadata endpoints (per-request, never cached in the index)
* Guardrail policy is readable by any tenant member, writable only by a tenant admin

## Auditability

* Immutable publish events record actor, digest, and policy verdict
* Retrieval events record a hashed caller identity and skill reference
* Publish events additionally record `guardrail_warning_ids` — which non-blocking checks fired, even though they didn't stop the publish — so a tenant can audit warning trends from the log alone